<div>Thanks a lot for all the help! Now I have a much better understanding on this issue.</div>
<div> </div>
<div>Best,</div>
<div>-Peter<br><br>********************************************<br>Peter Winey, Ph.D., DuPont<br><a href="mailto:peterwiney@gmail.com">peterwiney@gmail.com</a><br>******************************************** <br><br> </div>
<div><span class="gmail_quote">On 7/22/08, <b class="gmail_sendername">Stefano de Gironcoli</b> <<a href="mailto:degironc@sissa.it">degironc@sissa.it</a>> wrote:</span>
<blockquote class="gmail_quote" style="PADDING-LEFT: 1ex; MARGIN: 0px 0px 0px 0.8ex; BORDER-LEFT: #ccc 1px solid">Dear Peter Winey,<br><br>I think the problem with the elf + US is that the local kinetic energy in the core region is not well defined.<br>
In the PAW formalism the all-eletron wave functions can be reconstructed and from them the kinetic energy density could be calculated hence this is a well defined object, but in the US case you only have the augmentation charges that are insufficient to define it.<br>
In practice, I guess, the choice of NOT adding any augmentation kinetic energy is made which is somehow unsatisfactory (hence the "not fully implemented" cavet)..<br>Notice that even for NC pseudopotential the elf is unphisical in the core region<br>
since the core contribution should in principle also be considered, which is not.... not to mentin that the pseudowfc in the core are wrong...<br>So in conlusion I think that in general one should trust the calculated elf in the interstitial region but NOT in the core region (both for US and NC), only PAW, or some PAW-like wfc reconstruction (including core contributions) could give good results in the regions.<br>
<br>stefan
<div><span class="e" id="q_11b4b6d260b47956_1"><br></span></div></blockquote></div>--